03_score_group_analysis

Define Large Subreddits

large_subreddits <- pronoun_use_by_score_group %>%
  filter(
    score_group == "z1000+"
  ) %>% select(subreddit)

Error in eval(lhs, parent, parent) : 
  object 'pronoun_use_by_score_group' not found

Density plots of pronoun use

library(plotly)
plot_score_group = function (a,pronoun) {
  ggplotly(ggplot(a,aes_string(x = paste("contains_",tolower(pronoun),sep=""),color = "score_group")) + geom_density()+
             facet_wrap(~topic_level1)
             )}
large_sub_pronoun_use <- pronoun_use_by_score_group %>%
  filter(
    !subreddit %in% large_subreddits
  ) 
plot_score_group(large_sub_pronoun_use,"I") 

Groups with fewer than two data points have been dropped.Groups with fewer than two data points have been dropped.Groups with fewer than two data points have been dropped.

plot_score_group(large_sub_pronoun_use,"We") 

Groups with fewer than two data points have been dropped.Groups with fewer than two data points have been dropped.Groups with fewer than two data points have been dropped.

plot_score_group(large_sub_pronoun_use,"They")

Groups with fewer than two data points have been dropped.Groups with fewer than two data points have been dropped.Groups with fewer than two data points have been dropped.

agg_pronoun_use <- pronoun_use_by_score_group %>%
  filter(grepl("Sports", all_topics)) %>%
  group_by(score_group,subreddit,all_topics) %>%
  summarise(
    agg_i      = sum(num_comments*ZNcontains_i     )/sum(num_comments),
    agg_me     = sum(num_comments*ZNcontains_me    )/sum(num_comments),
    agg_we     = sum(num_comments*ZNcontains_we    )/sum(num_comments),
    agg_us     = sum(num_comments*ZNcontains_us    )/sum(num_comments),
    agg_they   = sum(num_comments*ZNcontains_they  )/sum(num_comments),
    agg_them   = sum(num_comments*ZNcontains_them  )/sum(num_comments),
    agg_you    = sum(num_comments*ZNcontains_you   )/sum(num_comments),
    
    total_comments = sum(num_comments),
    
    unweighted_i      =  mean(ZNcontains_i   ),
    unweighted_me     =  mean(ZNcontains_me  ),
    unweighted_we     =  mean(ZNcontains_we  ),
    unweighted_us     =  mean(ZNcontains_us  ),
    unweighted_they   =  mean(ZNcontains_they),
    unweighted_them   =  mean(ZNcontains_them),
    unweighted_you    =  mean(ZNcontains_you )
            
    
  )# %>% arrange(t1_t2) 
plot_categories_score_groups = function(agg_pronoun_use,pronoun){
ggplotly(
  ggplot(agg_pronoun_use, aes_string(x = "score_group", y = paste("agg_",tolower(pronoun),sep=""), name = "subreddit", color = "all_topics")) +
  geom_point(aes(size = total_comments))
)
  }
plot_categories_score_groups(agg_pronoun_use,"I")

plot_categories_score_groups(agg_pronoun_use,"Me")

plot_categories_score_groups(agg_pronoun_use,"We")

plot_categories_score_groups(agg_pronoun_use,"Us")

plot_categories_score_groups(agg_pronoun_use,"They")

plot_categories_score_groups(agg_pronoun_use,"Them")

plot_categories_score_groups(agg_pronoun_use,"You")

Analysis by score (not group)

Base rates across subreddits

#for pronoun in c("individual", "collective")
ggplotly(
  ggplot(pronoun_source, aes_string(x = "score", y = "ZNcollective",
                                                 name = "subreddit")) + geom_smooth(aes(size = num_comments))
  )

`geom_smooth()` using method = 'loess' and formula 'y ~ x'

large subs

large_subreddits <- collective_pronoun_by_score %>%
  filter(
    score >1000, num_comments > 10
  ) %>% select(subreddit)
large_subreddits <- large_subreddits[,1,drop = TRUE]

POT6

Size of subreddit vs. score-pronoun relation

collective_pronoun_by_score %>%
  filter(score == 50) %>%
  ggplot(aes(num_comments,ZNcollective))+geom_point()+
  #geom_smooth()+
  scale_x_log10()+scale_y_continuous(limits = c(-100,250))+ facet_wrap(~ topic_level1)